Are your tried-and-tested models not delivering the expected outcomes? Are you finding some elements of your strategy are just not working out? Over the last few years, many forward-looking companies have been setting and implementing data strategies. This has been based on the proof that those who do invest and excel in data have seen significantly higher profitability and success than those who did not.
The advent of Covid-19 has only served to exaggerate the gap between the data haves and have-nots, even in relatively unaffected sectors. But even those who can claim to be leaders are now experiencing difficulties and a gap between their expectations and reality.
So, what has changed? Firstly, in many cases, the most valuable/useful data a company had to answer business-critical questions was internal - customer, employee, process and systems data. These enabled an organisation to understand how it worked, who did what, how customers behaved. Using these data sources, analysts and data scientists have been able to drive decisions, advise on actions to take and predict the future.
Models would identify sales opportunities, segment customers, predict churn, identify cross-sell and up-sell opportunities, forecast campaign response rates and model the risk of non-payment. All of this was based on previous behaviour, blended with business knowledge, competitive intelligence and real-time contextual data.
Customer behaviour has changed or been changed
But something has gone wrong. Customers’ past behaviour and buying habits can no longer be relied on to predict what they will do next. In the grocery sector, for example, the big weekly shop is back and people buy for vulnerable others outside of their household. Meanwhile, product ranges and selection options have been reduced to protect the supply chain. In non-food retail, people who have never shopped online before are now part of a major upshift in digital buying. Leisure and transport has been turned on its head - who knows when we will be able to fly abroad safely for a holiday?
All of this has a major impact on the main attributes that featured in reports, analysis, models or segmentations, such as:
Think how much these features of existing models have changed over the last three months. Some 8.4 million consumers who had a decent job with good wages are now on Government furlough support and may end up on income support when this scheme tapers off. “Digital nevers” have had their eyes opened to a new world due to enforced isolation, using online shopping, Zoom and WhatsApp for the first time and may keep this behaviour post lockdown.
Pre-Covid-19 models are no longer valid.
The algorithms and models learnt what was going on based on historic data, either independently or assisted by an analyst or data scientist. These data sets, the key features they highlighted and therefore the outcomes they predicted are in many cases no longer valid. Even the hypotheses the work was based on are now changing as behavioural patterns shift, businesses pivot, processes change and the priorities of yesterday are irrelevant.
So, if the past behaviour and a lot of your own internal historic data won’t give the answers you need at the moment, what should you be looking at?
External data sets need to become part of the answer, blended with more recent or real-time transaction and interaction data. Potential sources include:
Rich external and open data sources can further enhance models, such as:
Will consumers rush back to pre-lockdown behaviours?
New behavioural modelling will be essential to see which habits endure post-lockdown and what were just passing trends. Which of your customers and employees will be rushing out of the door to embrace the experiences and interactions they have missed during lockdown and which will be scared or unable to go out due to severe anxiety or concerns over underlying health issues?
Nearer to home, companies will need to increase reliance on qualitative and quantitative research, coupled with near-real-time data from websites, searches or till transactions to check what people are actually doing. How will you get the necessary data (with sufficient history and stability) to re-segment and re-personalise your offerings to meet these new needs? Will customers be willing to share the personal information that would enable you to do this?
Then come the critical questions:
Can you react and pivot your data strategy with agility?
Do you have the digital and data capabilities to meet changing needs?
Do you have the data or can you source and integrate it quickly?
Do you have the right people to analyse, interpret and understand it and, if so, are they furloughed? When is data good enough to use for trends as opposed to perfect?
Do you have rapid A-B testing to understand quickly what is working and what isn’t?
Can you legally hold and use the data in line with GDPR and does the use of the data match your company values?
After companies and boards have been through the initial focus on finances, cash, liquidity and capital, they will start to look around more broadly on what to do next. Aside from the mental and physical wellbeing of employees and customers, next on their priority list will be having the right digital and data strategy and capabilities to ride out this storm.
James Morgan, CognitiveAnalytics